Handbook

nffi"H !--tA--

,o irllhH'tslSRH\lFtrsl\'ii =l

I "----1

Sign

ln this handbook, the concept of sign is gener-ally used in its broadest sense of a natural orconventional semiotic entiry consisting of a srgnyehicle connected with meaning. Many narrowerdefinitions o[the term sign have been given dur-ing the history of semiotics (cf. Tlpology f .) Themosr important models of the sign are discussedin the chapters on the classics of semiotics(Peirce, Saussure, Morris, and Hjelmslev). Inthis chapter, a synopsis of these and other signmodels will be given based on the standard dis-tinction berween dyadic and triadic sign mod-els. A few theoret ical and terminologicalpreliminaries deal with various distinctions be-rween signs in the context of related semioticand nonsemiotic phenomena. Two dimensionsof the sign, namely, sense and reference, are dis-cussed in the chapter on meaning. For furtheraspects, see Typology and Arbitrariness.

L Preliminary Distinctions andTheoretical Foundations

The defi.nition of rhe sign begins with problemsot terminology and the ontological quesrion ofthe naure of the sign and irs signifier as op-posed to rhe nonsemiotic world. These are ter-minological and theoretical preliminaries ro arypology of sign models.

| . I Terminological Preliminaries

There is a considerable rerminological vague-ness in the distinction between the sign, its sig-nifier, and its minimal elements.

I,I.I THE SIGN IS NOT THE SIGN VEHICLE

The sign is more than its constituent sign vehicle(cf. Peirce's representnmen, Saussure's signrf.er),but this distinction is often neglected. In every-day language, there are no words to distinguishbetween sign vehicle and the sign. The word srgnis ambiguous. It has either the broader sense of asemiotic entiry which unites a sign vehicle with ameaning, or it has the narrower sense of a signvehicle only. Both senses are probably impliedwhen we nlk about "raffrc signs." (The GermanIanguage has rwo words to distinguish berweenthe sign vehicle lVerhehrsschildl and the sign

lVerhehrsTeichenl.) The narrower definition refer-ring only to the sign vehicle is given inWebster'sThird International Dictionary: "Sign is a verygeneral term for any indication to be perceivedby the senses or reason."

In semiotics, the distinction between signvehicle and sign was introduced in various ter-minological versions (see the s)rnopses in figs.Si I and 3). But because of the ordinary lan-guage usage, this distinction has never beenstrictly observed. Even Saussure and Peirce didnot consistently distinguish between signifierand sign or representamen and sign.

I. PRELIMINARY DISTINCTIONS AND THEORETICAL FOUNDATIONS I 79

I.I.2 THE SIGN IS NOT A PHYSICAL SIGNAL

ln information theory, the term signal corres-ponds to the srgn vehicle of semiodcs (cf. Com-munication 3.1.2). This signal or informationvehicle (cf. Nauta 1972:282,294)is opposedtothe sigrr since it is oniy its phlisical embodiment.According to Klaus's Dictionary oJ Cybernetics,"signals are only potential sign vehicles. Insofaras they fulfrl the function o[ signs, this tran-scends their physical properries. Only those sig-nals are signs which transmit a message" (1969:569, 721). ln linguistics, Hockett adopted theterm speech signal rc characterize the linguisticsignifier in its physical form (1958: ll5). RoyHarris, in his translation of Saussure (1916e),translates signrfant as signal. For other defrni-dons o[ the signal, see Typologr of Signs (a.).

I.I.3 THE SIGN IS NOT A SIGN ELEMENT

In sign systems with a second level of articula-tion (see Language 4.1), the elements of (theminimal) signs are not signs themselves. Pho-nemes, for example, are not signs since theymean nothing. Terminologically, this differ-ence is neglected in definitions which extendthe term sign to include nonsigni$zing sign ele-ments. Resnikow, for example, states: "A ma-terial object which has no sign function of itsown, being only one o[ the eiements in the pro-cess of designation, is usually also called sign.ln this sense, the term 'sign' is also used withrespect to phonemes or letters" (I96a: If l.

This use of sign as a term for sign elementhaseven been adopted in the German industrialnorm DIN 44 300, which gives terminologicalrecommendations in the field of informationtechnology. It defines sign only as the minimalelement of a sign repertoire, such as letters, ci-phers, etc., and introduces the term symbolforthe meaningful units of the message (which aresigns in the terminology adopted here). Forother terminological distinc tions be tween signand symbol, see also Typology of Signs (3.1).

7.2 Ontolog of the Sign

What is the mode of existence of the sign andits signifier? Do they have a reai existence in

80 I SIGN

the form of a physical object or event, or dothey exist only as the perceptum in an act ofsemiosis? While this paragraph focuses on rheontology of the signifier, rhe quesdon of theontology of sense and reference wil l be re-sumed in the excursus on realism, conceprual-ism, and nominalism in semiotics.

I.2.I THE NATURE OF THE SIGN VEHICLEThe signifier of the sign has been characterizedas a concrete object, an abstract entiry, or borh(cf. Pelc l98la: 2-3). Morris defined the signvehicle as "a particular physical event or ob-ject" (I946: 96,367). Saussure had the men-tal ist concepr of the s igni f ier as a"psychological imprint. " Hjelmslev differenti-ated with respect to the sign vehicle further be-tween a physical and physiological expression-substance, whose study should be the subject ofphysics and psychology, and an expression-

/orm, which is the conceptional structure of rhesignifier.

In Peirce's semiotics, the sign vehicle orrepresentamen is either a concrete object, aperceptum, or an idea or "thought." ln one ofhis definitions, where he neglected the termi-nological d ist inct ion between sign andrepresentamen, Peirce defined the sign vehicleas "an Object perceptible, or only imaginable,or even unimaginable in one sense-for theword fasf,' which is a Sign, is not imaginable,since it is not this word itself that can be setdown on paper or pronounced, but only an in-stance of i t " ($ 2.230; cf . 1.2.3).

1.2.2 THE SIGN rS NOT A CLASS OF SEMTOTTC OBJECTS

In everyday language, there is a tendency roidentify signs (as opposed to nonsigns) with aclass of prototypical signs. These are neveremployed except as signs and belong to thesign repertoire of a code (cf. the technologicaldefinit ion discussed in f . i .3). Words, charac-ters, or conventional gestures are such proto-typical signs which can be listed as the signrepertoire or lexicon of a code. Againsr thisview, Morris objected that "semiotic is notconcerned with the sudy of a particular kindof object, but with ordinary objecrs in so far

W|E:r:.'

g:ti::'i;,:,g.L

F.Fr,.l ' -

Lfli ,f'l r

iII

i

171nd only in so far) as they participate in

semiosis" (I938: 4). Every object, event, or'behavior is thus a potendal sign. Even silence

can have the semiotic function of a zero sign

(cf . Sebeok L976: Ll8, Chronemics 4.2. i ) .

Everything can thus be perceived as a natural

sign of something else, and by prior agree-

ment between a sender and a receiver, every

obiect can also serve as a conventional sign.

This does not mean that every phenomenon

of the world is semiotic. It only means that

under conditions of semiosis every object can

become a sign to a given interpreter.

I.2.3 ACTUAL AND POTENTIAL SIGNS, TOKEN, AND

TYPE

I[, as Morris argued, "something is a sign onlybecause it is interyreted as a sign of somethingby some interpreter" (1938: 4), it must be con-cluded that signs cease to exist as signs whenno interpreter perceives them. Does this meanthat words in a lexicon or the characters of anextinct language are usually nonsigns? Two se-miotic concepts have been developed to avoida simple negative answer to this question, typeand potential sign.

Kamlah & Lorenzen distinguish between ac-tual and potential signs (1967: 58). Words in aIexicon which are acually not read can thus bedescribed as potential signs. In the world ofnatural semiosis, the number of potential signsis unlimited (cf. 1.2.I). When signs belong to acode, there are as many potential signs as thereare elements in the sign repertoire of the code.Peirce introduced the distinction between to-hen and type. Asign in its singular occurrence isa token, whereas the sign as a general law orrule underlying its use is a t1pe. An example isthe word/ast in the passage from Peirce quotedabove (1.2. I ) As a word of the Engl ishlanguage it is a type. Every written or spokeninstance of it is a token. The linguistic dimen-sions corresponding to the token-type dichot-omy are text and system. In terms of thesesemiotic categories, signs do not exist only astokens in actual processes of semiosis. Theyalso exist as types, as the user's semiotic poten-tial of lawful sign use.

1.3 The Sign and the Nonsemiotic World

The nonsemiotic world is related to the sphereof semiotics in two ways. One has to do with

the process of reference. The other has to dowirh the delimitation of signs from nonsigns.Before the definitions of the sign are discussed,an oudine of the main views of semiodcians onthe nature of the nonsemiotic world will begiven. There seem to be six main approachesto this problem, transsemiotic agnosticism, pan-semiotism, naive realism, pragmatic mediational-ism, functio nalism, and inte gr ativ e ho lism. Someof these approaches have parallels in the semi-otic approaches to the problem of reference(see Meaning 2.).

I.3 TMNSSEMIOTIC AGNOSTICISM AND

PANSEMIOTISM

Orthodox structuralism defends a transsemi-otic agnosticism- According to Saussure, thenonsemiotic world is "a vague, uncharted neb-ula." Since "nothing is distinct before the ap-pearance oflanguage" (Saussure I9I6b: III-I2), nothing can be said about the nonsemioticworld. Pansemiotism seems to maintain theopposite view: the whole world is a semioticsphere. Peirce is the crown witness of this ap-proach: "The entireuniverse [ . . . ] isperfusedwirh signs, if it is not composed exclusively ofsigns" ($ 5.448, fn.). Under quite differentpremises, pansemiotism was also defended inmedieval theology (q.v. I .3, see also EcoI973b. I I I for pansemiot ic meraphysics).Variants of pansemiotism occur also in infor-mation theory (see Information 4.1.I: percep-tion as information), communication theory(see Communication 2.6.2: the metacommuni-cative axiom), semiotic epistemology (cogni-tion as semiosis; cf. Guti6rrez Lopez 1975),endosemiot ics (see Communicar ion 2.2.2).and semiotic views of molecular biology (cf.Prodi 1988 and the discussion of the geneticcode in Code 5.1).

The difference between pansemiotism andtranssemiotic agnosticism is not as fundamen-tal as it seems. Both refuse to assume a non-semiotic sphere. An attempt to bridge the gap

I. PRELIMINARY DISTINCTIONS AND THEORETICAL FOUNDAIIONS I 8T

between the two approaches is Greimas's the-ory of a natural semiotics (cf. Meaning I.3,2.3.2).In this theory, the nonsemiotic worldbecomes semiotized in a natural semiotics, andthe relation between the semiotic and thenonsemiot ic is reinterpreted in terms ofintersemioticity.

I,3.2 NONSIGNS IN NAIVE REALISM

St. Augustine represents a naive realism (see

below, 1.4) in his division of the world intothings and signs (res and signum) (397: 624-25). In De doctnna christiana I. 2, he gave thefollowing account of "What a Thing Is, andWhat a Sign: [ . . . ] I use the word'thing' in asrrict sense to signi$r that which is never em-ployed as a sign of anything else: for examplewood, stone, cattle or other things of thatkind." But Augustlne also knew that signs arenot a class of objects which is ontologically dis-tinct from things (cf. I.2.2): "Every sign is alsoa thing; for what is not a thing is nothing at all.Every thing, however, is not also a sign." Al-rhough logically separate, the spheres of thingsand signs are not epistemologically unrelated,for according to Augustine, "things are learntby means of signs."

I.3,3 NONSIGNS IN MEDI,ATIONAL AND zuNCTIONAL

THEORIES OF ACTION

lmplicitly or explicitly, the theories of semioticmediation (cf. Merz 6z Parmentier, eds. 1985)characterize human interaction with the non-semiotic world from the point of view of a gen-eral theory of perception or behavior. The keyto the difference between signs and nonsigns isthe dichotomy of mediated vs. nonmediatedpercepdon. Whitehead characterizes the worldbelow the "semiotic threshold" as a sphere ofperceptive immediacy: "The immediate worldaround us [is] a world decorated by sense-datadependent on the immediate states of relevantparts of our bodies. [ . . . ] 'Sense-datum' is amodern term: Hume uses the word 'impres-

sion.' " ln contrast to such "presentational im-mediacy," "the human mind is functioningsymbol ical ly when some components ofits experience elicit consciousness, belieG,

82 I SIGN

emotions, and usages, respecting other com-ponents of i ts exper ience" (1928: 16, 9).When Morris defined semiosis as "a mediated-taking-account-of" ( I938: 4), he also char-acter ized nonsemiot ic behavior as anunmediated interaction with objecrc of theworid.

One of the most explicit theories of semioticmediation has been proposed by Vygotsky(1930: 137-38; cf. Rissom 1979: Ll). Vygot-sky distinguishes two elemenmry forms of hu-man behavior ' . natural and art iJ ic ia l orinstrumental acts. In natural acts, there is a di-rect associative (conditioned reflex) connec-don between a stimulus A and a response B. lninstrumental acts, "two new connections, A-Xand B-X, are established with the help of thepsychological tool X." According to Vygotsky,such a tool X is a stimulus which functions "as

a means o[ influencing the mind and behavior"(ibid.: lal). In other words, the mediatingstimulus X is a sign, while direct acts are formsof nonsemiotic behavior.

In the fiamework of his functional stmctur-alism, Mukaiovsky also draws a dividing linebetween signs and nonsigns by means of thecriterion of immediacy of action (1942: 4l-42;cf. Function 3.3). Nonsemiotic behavior, in histheory, is the immediate (practical or theoreti-cal) interaction with reality, while a sign pre-supposes a mediated interaction, where thesign is the mediator between two realities.

I.3.4 THE PRESEMIOTIC SPHERES IN INTEGRATIVE

HOLISM

Some philosophers and semioticians have pro-posed holistic world models in which thesphere of signs is assigned a place beside anonsemiotic (or several such) world(s). Popper& Eccles's three worlds are a prominent exam-ple of such an approach (1977: I6ff.) . World I ,the world of physical objects, is clearly thenonsemiot ic sphere. Worlds 2 and 3 arespheres of increasing semioticity. World 2 isthe world of subjective experience, and World3 comprises the products of the human mind.ln the framework of his Evolutionary CulturalSemiotics, Koch (l9B6a; b; c) develops a holis-

dc world model in which the dividing linebetween the semiotic and the nonsemioticworlds is drawn according to evolutionaryprinciples. Within the presemiotic sphere, thefollowing five "worlds" are distinguished asevolutionary stages: (I) the cosmic, (2) the ga-lactic, (3) the geological, (a) the biological, and(5) the sociological world (I986b: I2).

1.4 The Typolog of Sign Models

Sign models can be classified according to sev-eral semiotic dimensions. Most of the criteriadiscussed in the context of the typology of theo-ries of meaning are also valid with respect to the

rypology of sign models. This chapter will sur-vey the major models of the sign on the basis ofthe distinction berween dyadic and riadic mod-els. Criteria of an altemadve classification arediscussed in the following paragraphs on real-ism, conceptualism, and nominalism in semiot-ics. These aspects of the typology of sign modelsare especially relevant to the history of semiot-ics. The framework provided by the philosophi-cal disdnction between realism and nominalismis coextensive neither with the dyadic-triadicdichotomy nor with the rypology of theories ofsense and reference outlined in the chapter onmeaning (but see Meaning 3. i). For interpreta-tions of sign models in terms of the nominalism-conceptualism-realism tradition, see Woozley(1967), Kutschera (I971: 3I-78, only on real-ism), Trabant (1976a:23-27, only on nominal-ism and realism), Lyons (1977: I09-I14), and

Jadacki (1986).

I.,',I DYADIC. TRIADIC. AND OTIIER MODELS OF THE

REI-ATA

A standard typology of sign models distin-guishes dyadic and triadic models on the basisof the number of relata characterizing the signin its semandc dimension (cf. Meaning). Tri-adic models distinguish between sign vehicle,sense, and reference as three relata of the sign.Dyadic models'ignore either the dimension ofreference or that of sense. Dyadic models aresometimes developed into tetradic models by afurther dyadic subdivision of the two into four

components (or planes) of the sign. Hjelm-slev's sign model is an example of such anextension of a dyad to a tetrad. Auroux's sug-gestion that most other semiotic dyads are es-sentially based on a tetrad (1979: 24) is notvery convincing.

The distinction between dyadic and triadicsign modelg has been interpreted as being bothfundamental and unbridgeable, although therehas not always been a clear-cut distinction be-tween these two t)?es of models in the historyof semiotics (cf. 2.1.i). Followers of the dyadictradition have elaborated on the aporias of auiadic theory of the sign (cf. Fischer-Lichte1979: 38-5I), while followers of the triadictradition have criticized the inadequacy of thedyadic model (cf. Koller 1977: 25-33).lt mustbe emphasized that the option for either one ofthe two models does not imply the neglect ofthe pragmatic dimension of semiosis (paceSchaff 1960: 205; cf. Meaning 4.). In any casethe interpreter is an additional relatum o[ thesign (see also Lieb I98Ia I44).

Are there sign models proposing less ormore than two or three relata of the sign? Amonadic view of rhe sign which neglects to dif-ferentiate between sign vehicle and meaningoccurs only outside of the theory of signs. It ischaracteristic of magic and unreflected modesof sign manipulation. General Semantics warnsagainst this view of the sign with the slogan"The symbol is not the thing symbolized"(Hayakawa l94l 27; cf. Semantics I .5). A te-tradic model of the linguistic sign has beenproposed by Hockett (1977:82), who arguesthat the triad of word-idea-thing should be ex-tended by an additional conceptual unit of the"image of the word." (Notice that Saussure'ssignifier is also a mentalist "acoustic image.")

In linguistic lexicology, K. Heger's semiotictrapezium is a sign model with more than threerelata. lt extends the classical triad to a tr^pe-zium which specifies the following six relata ofthe linguistic sign (Baldinger 1970: I55-56):( I ) phonic substqnce, (2) moneme (sum ofsememes and phonemes), (3) sense or signifed(sum of all meanings associated with one signi-fier), (a) sememe (one particular meaning), (5)

I. PRELIMINARY DISTINCTIONS AND THEORETICAL FOI.'NDAIIONS T 83

concept or seme (a ianguage-independent, ele-

menrary meaning), and (6) realiry, (thing). Syn-

optic surveys of dyadic and triadic sign models

from the history of semiotics are given by Eco(I973b: 30), Nattiez (1,979: 391), and Faltin

( leB5; 30).

I.4.2 REALIST MODELS OF THE SIGN

Semiotic realism in its most genuine form

originates from the philosophy of Plato (cf. His-

rcry 2.2.2), The correlates o[ the sign are as-

sumed to be nonmental entities. The extremerealist believes that both sense and reference

(cf. Meaning 2.-3.) exist in themselves andwould exist even if there were no minds to be

aware of them. In a mindless world, theywould be available for discovery, even if therewere nobody to discover them (cf. Woozley1967 L94-95). Plato, Aristotle, Augustine,Bolzano, and Frege are counted among therealists in the history of semiotics. For Hus-serl 's semantic realism, see Meaning (3.1).

I.4.3 CONCEPTUALIST MODELS OF THE SIGN

ln semiotic conceptualism, the semantic di-mension of the sign is assumed to be mind-de-pendent. For the conceprualist, meanings existin the mind in a subjective sense, such that if

rhere were no minds, there could be no mean-ings (cf. Woozley 1967: 195). Conceptualismraises the question of the referential correlateof the srgn and its reality only insofar as this"reality" provides the sense data to the mind.The main representatives o[ conceptualism arethe Bridsh empiricists Locke, Berkeley, andHume. John Stuart Mill and Franz Brentano aremodern (nineteenth-century) conceptualists inlogical semantics. See also Meaning (3.2) formentalism in modern semantics.

I.4.4 NOMINALIST MODELS OF THE SIGN

Semiodc nominalism rejects the idea of a real-ity of general concepts or referents (cf. History2.2.2). Nominalists acknowledge only the exis-tence ofsingular objecrc and deny the reality ofuniversals, i.e., the properry predicated of all

84 I SIGN

the individuals of a certain class, such as "red-

ness." In its extreme form, nominalism arguesthat objects having the same qualiry have noth-

ing in common but their name (Lat. nomen,therefore: nominalism). William of Ockham,for example, taught that universals are only

signs without an existence of their own, stand-ing for individual objects or sets of objects (c[

Geyer, ed. l95l 576 and Woozley L967:

203). These signs are thus only names without

any correlate in realiry.According to Jadacki, "recent tendencies to

'semiotize' all areas [ . . . ] concernedwith theformal or empirical aspects of meaning andreference, can be interpreted as expressing thefact that nominalism today is, more or less, thedominant school of thinking: everything con-cepual only exisrs in the use of its sign" (1986:II36). Modem semiotic nominalists, accord-ing to this interpretation, are thus the struc-turalists, and semioticians in the l ine fromSaussure to Hjelmslev, Greimas, and Eco.However, this modern semiotic nominalismtends ro be sdll more nominalist rhan rhe rradi-tional one which acknowledged at least the ref-erenrial reality of individuals.

2. Dyadic Models of the Sign

The most general dyadlc characterization ofthe sign is given in the medieval formula aii-quid stat pro aliquo, "something stands forsomething else." According to Eschbach(1980: 44), one of the first explicit quotationsof this formula is in the writings of Albert theGreat (thirteenth century), but the Scholasticsusually used a different formulation, supponitaliquid pro aliquo, "something serves in placeof something else" (cf. Kneale & Kneale I962:250). Definit ions of the sign using the aliquidpro aliquo formula may be dyadic or rriadic. Ifthe aliquo is subdivided into sense and refer-ence (see Meaning), the'definit ion is extendedfrom a dyadic to a triadic one.

2.1 Aliquid pro Aliquo

The representative function (stare prd of thesign has been a criterion of the definition of thesign from Augustine to Jakobson. What is thenature of the relata and of the reladon?

2.I.I DYAD OR TRIAD IN THE HISTORY OF SEMIOTICS

One of the earliest explicitly dyadic models ofthe sign was proposed by the Epicureans (seeHistory 2 I 4), who rejected the Stoic lehton asthe third correlate of semiosis. The aliquid proaliquo formula of medieval semiotics suggests adyadic model with the two correlates of thesign vehicle (aliquid) and its referent (oliquo).However, a closer look at the definitions oftenreveals reference to a third correlate. ln Augus-tine's definition (De doctrina christiana 2.I.1;c[ History 2.1.5), "a sign is a thing which, overand above the impression it makes on thesenses, causes something else to come into themind as a consequence of itself" (397: 636).Markus (1957:7I-72) and Simone (1972: 16)have interpreted this definidon of the sign as atriadic one, consisting of (I) the sign vehicle,(2) its referent, and (3) the mind to whom thesign stands for the object. However, the modelis not triadic in the uadition of a sysremaricdistinction berween sense and reference (c[.Meaning), as it was postulated by the Stoics(see 3.2). The focus is on the sign vehicle andrhe referent. The interpreter is a necessary pre-requisite of any sign model even though it maynot always be mentioned explicitly. The deci-sive difference between triadic and dyadicmodels remains the disdncdon between senseand reference, which seems to be of no con-cern to Augustine.

The Scholastics developed the thegry ofstarepro further in their theories of representation(cf. Kaczmarek I986) and supposition (seeHistory 2.2.3). Ockham's defi.nition of the signrs stili very similar to Augusdne's: "A sign isthat which makes something else come toone's mind" ("ille, quod aliquid facit in cogni-t ionem venire," quoted in Geyer, ed. I95l :578) Aimost the same definition is used byPornsot: "A s ign is something that makes

something other than itself present to knowl-edge" (1632:25).

The question whether rhis aliquid-aliquo dyadinvolves a semiodc dyad or a triad has been in-terpreted differendy in the course of the historyof medieval semiotics. Coseriu gives evidencefor a triadic conception in early medieval semi-otics consisting of the three terms vox-conceptio-res. He quotes Boethius (480-524) as fbllows:"The voice (vox) signifies the conceps (concep-tiones) of soul and intellect, but the same intel-lect conceives of the things (rCI) and is signifiedby the voices" (1970: 153). (See also History2.2.4.) ln early Scholastic semiotics, this Stoictriad was broken up into two dyads. Thus,Anselm of Canterbury (1033-II09) distin-guished two semantic relarions, signif.catio andappellotto (Pinborg 1977: 43): appellatio is therelation benveen word and thing, while signrfica-tio is the relation benveen word and its concep-fual content. ln the later development of thesemiotics of the Modistae, there was a shift fromthe triad to a word-thing dyad, although the me-diating role of the intellect was still acknowl-edged (cf Bursill-Hall I97T:97, Ebbesen I983:73, Gabler I987:'18).

2.1.2 ALIQUID PRO AUQUO IN MODERN SEMIOTICS

With reference to H. Gomperz, Btihler resumesthe oliquid pro aliquo formula as an element inhis theory of representation (1933b: 93-96).He describes rhe aliquid (the sign vehicle) as aconcretum which can stand for (function as arepresentative of) the aliquo only by a processof abstraction. Brihler calls this the principle ofabstractive releyance. Jakobson follows this tra-dition when he states, "Each and every sign isa reJerral (renvoi)" (I97 5: 22) . Even Peirce ac-cepts the aliquid pro aliquo formula, althoughhis general semiotic framework is triadic: "Asign, or representomen, is something whichstands to somebody for something in some re-spect or capaciry" (S 2.228).

2.I.3 THE R-EI.ATAThe aliquid pro aliquo formula is open to bothdyadic and triadic interpretations. The followingrela ta of the " stat pr o" -r elation are characteristic

2.'DYADIC MODELS OF THE SIGN T 85

of different rypes of sign models (cf. Meaning 2.-3. and Wiegand 1970: 249-5I): (I) signifier-re-ferent (the "nomen significat rem-" theory ofRoman grammarians; cf. Padley 1976:16fl, (2)(Saussure's) signif,er-signified (sense), (3) theunity of signifier plus signified-referent, (4) atriad as a coupling of rwo dyads (cf. 2.1.1 andLocke in 2.3), and (5) the (triadic; see 3.) signi-fier-signifi ed-referent model.

2.2 The Relation of "Standingfor"

Peirce's specification that the "standing [or"must be "for something in some respect" hasthe advantage of precluding a common misin-rerpretation. It says clearly that the reladon of"standing for" is not one of substitudon (cf.Weils 1977: 6, pace Morris 1946: 84). The moreprecise nature ofthe "standing for" reladon de-pends, ofcourse, on the nature ofthe relata. Theclass of narural signs, where the interpreter re-lates the sign vehicle (rhe sympmm or index) toits referent by means of an inference, raises thequestion whether it is still appropriate to call thisrelation one of "smndrng for."

2.2. I SIGNIFICATION AND DESIGNANON

lf. the aliquo is the referent, the relation is oneof.designation or reference; ifit is sense, the rela-tion is one of signifcation (see Meaning 1.2).The medieval disdnction was between oppella-tio and signifcatio.lf the signifier refers to thereferent via the relatum o[ sense, this triadic re-ladonis oneof mediotion (see L3.3). Forrepre-sentation as a semiotic relation, see Meaning(1.2.3). The nature oI rhe aliquid-aliquo rela-rion can furthermore be specified asbeingarbi-trary (symbolic) or motivated (iconic orindexical; cf. Peirce 3.2).

ln the framework of their triadic sign model,Ogden 6s Richards specify the relation of thesign vehicle ("symbol") and the referent as be-ing without any "relevant relation other thanan indirect one," while they see relations ofpsychological or social "causality" betweenthe relata sign vehicle-sense ("thought or refer-ence") and sense-referent (1923: 10-lt). To-day, the concept of causality is certainly

86 I SIGN

inappropriate in this context. A modern suc-cessor is the concept o[ inference.

2.2.2TIiE DYAD IN NATUML SIGNS

According to Augustine's definidon of the natu-ral sign, nanrral signs are the object ofunilateralobservation, in contradistincrion to signs in bi-iareral communication. The sign vehicle (A) andthe referent (B) in this rype of narural semiosisare related in two ways. At an extrasemioticIevel, the level of narural evens, A is the effect o[rhe cause B. At the semiotic level, the effect Abecomes an index or qrmptom which an inter-preter connects by inference to B. Hobbes ex-plains the genesis of narural srgns (and moreoverof signs in general) as a process of interprenngthe consequent as an effect o[ its antecedent orvice versa:

When a man hath so oJten observed Iike anteced-ents to be followed by l ike consequents, thatwhensoever he seeth the antecedent, he lookethagain for the consequent; or when he seeth theconsequent, maketh account there hath been theIike antecedent; then he calleth both the anteced-ent and the consequent signs of one another, asclouds are signs of rain to come, and rain ofclouds past. Q6a0: a.9)

Larer, Wolffwas ro divide rhis field of naruralsigns into demonstrative, prognostic, andrememorative signs depending whether the re-ferent is a present, future, or past event (I720:

$$ 953-5a; cf. History 3.4.I). For a more re-cent theory ofnatural signs, see Clarke (1987)

2.2.3 INFERENCE AND SEMIOSIS

Is the inference by which an interpreter con-nects the antecedent with its consequent a signrelation (ct Pelc I9B4b)? Critics of a broad con-cept of the sign have argued that clouds do notmeanrain and that therefore the distincdon be-tween indication and signification, indices andslrmbols is a fundamental threshold from non-signs to signs (cf. Savigny 1974: 1788, Clarke1987). From a pragmatic point of view, Alstonargues that "there is a sharp disdnction betweenthe two groups of facrc. One is a matter of cenainde facto correlations holding and not a matter o[

the x's being used in a cenain way. The other is amatter of the way the x in question is used andnot a matter of correlations in which it standswith the y" (1967: +aQ . Clarke proposes to ex-clude inferences from narural evidence from thedefinition of signs (1987:49-50): "Clouds maysignify rain [ . . . ], but they clearly do not referto anything at all. To assign them a reference tothe rain is to collapse the crucial distinction be-rween the significance and reference of a sign."According to his own proposal, a natural signshould "be defined as an event having signifi-cance for an interpreter which is not producedfor the purpose of communicadon and whoseinterpretation does not require an inferencefrom a linguistic generalizadon." This proposalis an attempt to mediate between two ancientrival theories of the sign, those of Stoic and Epi-curean semiotics.

The Epicureans defended a realist view ofthe natural sign. According to Sextus Em-piricus (Against the Loglcians lI: 269-71), natu-ral s igns can be apprehended by directobservadon and do not require the mediationof language. Even animals can thus interpretnatural signs. This view of natural semiosis hasbeen accepted in the pansemiotic traditionfrom Augustine to Peirce and Morris.

An early nominalist view of narural semiosiswas the Stoic theory of signs. The nominalistEco (1976: 17; I984b:31) sympathizes withthis view because it makes natural semiosis de-pend on a criterion of "culrural recognition."According to the Stoics, the natural sign is notthe association between a natural event A("smoke") and a physically related event B("fire"), but the result ofa process ofhypothet-ical reasoning. lnstead ofbeing events, A and Bbecome propositions. The sign A is the ante-cedent of a hypothetical reasoning ("If there issmoke . . . ") and B is its inferred consequent(' . then there must be fire"). This rational-ist view of natural semiosis is untenable withinthe larger framework of general semiotics. Itsconsequence would be the exclusion ofzoosemiot ics and large sectors of anthro-posemiotics (for example, the semiotics of mu-sic') from the semiotic field.

2.3 The Dyadic Sign: A S;r'nopsis

A clear-cut distinction between dyadic and tri-adic sign models is not always possible (seealso 2. I . l). There is a zone of vagueness when-ever a third correlate is mentioned but not sys-tematically incorporated into the semiotictheory. Saussure's model is the prototlpe of adyadic model. Although he mentions the"chose" in addidon to the signifier and the sig-nified, he rejects it as a third correlate of thesign. Hjelmslev, while discussing reference inthe domain of content-substance, also remainsessendally dyadic in his concept of the sign.Furthermore, there are semioticians who pos-tulate two aspects of the sign but consider therelation between sign vehicle and meaning tobe the third component.

Major dyadic definitions of the sign from thehistory of semiotics are summarized in the s1m-opsis of Figure Si I. The definitions on whichthis suwey is based are as follows:

I. Augusdne, see 1.3.2, 2.1.I, and History(2 r 5).

2. Alberrus Magnus and the Scholastics, see2.1.r .

3. Hobbes, see2.2.2 and History Q.2.2).4. Locke (1690) distinguishes two types of

signs, (l) ideas as signs of things and (II) wordsas signs of ideas (see also Meaning 3.2): (I)"For, since the things the mind contemplatesare none of them, besides itself, present to theunderstanding, it is necessary that somethingelse, as a sign or representation of the thing itconsiders, should be present to it: and theseare ideas" (Locke 1690: IV.2I.4). (II) "Words

[ . . . ] stand for nothing but the ideas in themind oJ him that uses them. [ . . . ] That thenwhich words are the marks of are the ideas ofthe speaker" (Locke 1690: 111.2.2). See alsoHistory (3.3.3).

5. Port-Royal (: Arnauld & Nicole 1685),see Rey (1973: I l2), Auroux (I979:22), Swig-gers (I981), and History (3.1.2).

6. Wolff (1720 S 293; cf. Coseriu 1967:98and Trabant 1976a l7): "Thus, if two thingsoccur always simultaneously or one always af-ter the other, then one is always a sign of the

2. DYADIC MODELS OF THE SIGN . 87

orher. Such are called natural signs." See alsoHistory (3.1.1).

7. Degrirando (1800: I , 63), cf . History(3.3.4).

8.-9. See Saussure and Hjelmslev.10. Cassirer (1922-38 175; cf . Krois

1984b: 440): "Under the term symbolic formshould be understood each energy of humanspirir through which an intelligible content andmeaning is joined with and internally adaptedto a concrete sensible sign." On Cassirer's the-ory of the sign and symbol, see History Q.6.2)and Typology (L.2.2).

Il. Brihler (1933b; 1934), cf..2.1.2.12. Bloomfield (I933), cf. Meaning (4.1) on

behaviorism in semiotics.

I3. Buyssens (1943 lZ,34-4I), c[ System(4 2)

14. Jakobson ( I959a: 260; L96l : 575;

1975: 10ff.; cf. Waugh L976:38-53).15. Goodman (1968: xi, 5) does not differen-

tiare between sign and sign vehicle. His term

rymbol "covers letters, words, rexts, pictures,

diagrams, maps, models, and more." His refer-

ential view of the "s)'mbol" is apparent in this

quote: "The plain fact is that a picture, to repre-

sent an object [footnote: I use 'object' indiffer-

ently for anything a picrure represents, whether

an apple or a battle], must be a symbol for it,

sund for it, refer to it." See also Image (3.4).

Goodman's r epr esen tation, descrip ti o n, exem'

plifcation, and expression (1968: 256) are not

Sig.Correlates o

srsnPehicle I

the Dyad

@meaning

(I) Augustine (397) srgn sign (as thing) (other) thing(s),something else

(2) Albenus Magnus andScholastics (l3th cent.)

srgnum aliquid (vox) aliquo (res)

(3) Hobbes (r6a0) srgn antecedent experience consequent experience

(a) Locke (1690) srCn (1)sign (2)

ideaword

thingidea

(5) Port-Royal (Arnauld &Nicole 1685)

sign idea of the representingthing

idea of the thingrepresented

(6) woH (1720) srgn one thing another thing

(7) Degerando (1800) slgn sensatron idea

(8) Saussure sign(signe)

signifier(signifiant)

signified(sigrifie)

(9) Hjelmslev srglr exPressron content

(10) Cassirer (I923tr.) symbol(ic) form concrete sensible sign content, meaning

(lr) Biihler (I933b) slgn rePresentative (concretething)

meanrng

(I2) Bloomfield (1933) linguistic form speech sound, signal response in hearer

(13) Buyssens (1943) seme semrc act meaning, signification

(14) Jakobson (1959tr.) sign(um) signans signatum

(r5) Goodman (1968) symbol [words, pictures, models,etc. l

denotatum, object

Fig. 5i I. Synopsis of dyadic models of the sign. (See alsoSemiotics.)

88 I SIGN

the synopsis M 3 on meaning, and History o[

variants of sense (seeclaims (1985: 30), butfunction.

Meaning 3.), as Faltinfour tlpes of referential

3. Ttiadic Models of the Sign

Triadic sign models comprise a nonhomoge-neous group of semiotic theories distinguishingthree correlates ofthe sign, sign vehicle, sense,and referent (cf. Meaning). ln some cases,there is no clear-cut disdnction between dy-adic and triadic models (cf. 2.1.1). For surveysof triadic sign models see Gomperz (f 908: 76-9I) and Lieb ( l98la).

3.7 Some Types of Triailic Sign Moilek

The nonhomogeneous character of these signmodels can be illustrated in a tentative typol-ogy of triadic models. The basic disdncdon willbe berween triads that are reducible to dyadsand genuine triads.

3.I.T TRIADS REDUCIBLE TO DYADSSome riadic sign models are actually reducibleto rwo dyads. These may be either subsequentor alternative dyads. Locke's definition impliestwo subsequent but still potentially indepen-dent dyads: words are signs of ideas and ideasare signs of things. Anselm's distinction be-tween signlf catio, the relation between wordand concept, and appellatio, the relation be-nveen word and thing, implies two alternativedyads. In modem semantics, too, the distinc-tion between sense and reference is sometimestaken to be a matter of alternative dyads. Thisis the theory that words have either sense orreference (cf. Meaning I.). Others have postu-lated a genuine triad claiming that there is al-ways some sense and reference in signs.

3.I.2 GENUINE TRIADS AND THE SEMIOTIC TRIANGLE

Genuine triads are based on the concept of me-diation (cf. I.3.3): a third correlate is related toa first via a second. After Gomperz (1908 77),Ogden & Richards (1923: I l) have repre-

sented the triadic structure of the sign bymeans of a triangle. This diagram (cf. Fig. Si 2)has become known as the semiotic triangle (cf.Lyons 1977: 96, Lieb l98fa). It shows thethree correlates ofthe sign in the order (I) signvehicle, (2) sense, and (3) referent (Ogden &Richards use different terms, cf. synopsis inFig. S 3). The domed base line indicates the in-direct nature of the relationship between thesign vehicle and the referent and thus the pathof mediation from (l) to (3). However, the or-der of the relata in the process of triadic media-tion has been interpreted in different ways.

SIGNVEXICLE

Fio

REFERENT

Aristotle's definition of words as signs of thesoul, and the latter as l ikenesses of actualthings, gives the outline of the standard orderofthe triad: (l) signvehicle-(2) sense-(3) refer-ent (Fig. Si 2). Sense is the mediator o[ the re-ferent. ln medieval semiotics: "Vox significatrem mediantibus conceptibus" (cf. Lyons1977: 96). This is also the order of Ogden &Richards's triad and of Vygotsky's mediationalview of the sign. Peirce's definition of the sign($ 2.228) enumerates the three relatarepresentamen-interpretant-object in thesame order, but within his categorial system,the object is a phenomenon ofsecondness, andthe interpretant is one of thirdness.

In contradistinction to this smndard order ofthe triad, Plato's and Aristotle's sign modelsalso suggest a different interpretation of the se-quence of the relata (cf. Schmidt 1969: I3).Since in the Platonic radition, ideas are "like-

nesses" of acual things, this early picture the-ory of meaning (cf. Image 2.2) assigns theobject in a way the first place within the triad:

3. TRIADIC MODELS OF THE SIGN I 89

5i 2. The semiotic triangle (cf. text).

rhe thing (l) evokes the idea (2) which namesthe word (3). A still different order is suggestedby Husserl's phenomenology of meaning (q.v.3 I.3) Husserl's theory of the "meaning en-dowing act," in connection with his Platonicview of sense, assigns the first place within thesemiotic triad to sense: sense (l) "gives life" toa "name-Thing" (2) which refers to the object(3). More generally, the order (I) sense-(2)sign vehicle-(3) object is the order of semiosisfrom the point of view of sign production,while the order (l) sign vehicle-(2) sense-(3)referent is the order of semiosis from the pointof view of the interpreter.

3.2 The Triailic Sign: A Synopsis

The semiotic riangle (Fig. Si 2) is the frame-work of the following synopsis of major triadicdefinitions of the sign (Fig. Si 3). The synopsisis based on the following definidons:

l. Plato: Crotylus (cf. Arbitrariness 1.1.l,Meaning 3.1.I, Gomperz 1908: 79, Schmidt1969: 19, Coseriu 1970: 46, Schmitter 1987:28,32). The Greek terms are name-6vopo,

v6poq, idea, notion-si6oq, l.6yog, 6to-v6r1po, sound, voice-g06yyoq, e06Tpo, andthing-rpdypo, ouolo.

2. Aristode. De interpretatione (I6a tans.

J. L. Ackrile; cf. Lieb l9BIa): "Now spokensounds (q<ovtt) are symbols (otippolo) of af-fections (n&O1pcrc) in the soul (ryulfr), andwritten marks symbols of spoken sounds. Andjust as written marks are nor rhe same for allmen, neither are spoken sounds. But whatthese are in the first place signs of (oqpeio)-affections of the soul-are the same for all; andwhat these affections are likenesses of-actualthings (rpdylloro)-are also the same."

3. The Stoic sign model is described by Sex-tus Empiricus (Adtersus mathematicos 8.1l-2;cf. Krezmann 1967: 364): "The Stoics [ . . . Isaid that rhree things are linked togerher: (l)what is conveyed by the linguistic sign (cdorlpctv6pevov) (2) rhe linguistic sign itself(td oqpoivov) and (3) the object or evenr (tdtvy26d,vov) t I Two of rhese are corpo-real-viz. the sound and the objecr or evenr-and one is incorporeal-viz. rhe matter ofdiscourse conveyed by the linguistic sign, rhe

SignCorrelates of the Ttiz

ot@sign vehicle I sense

rd@

referent

(I) Plato (ca. 400 r.c.) name sound idea, content thing

(2) Aristode (ca. 350B.c.)

[sign] sound affections thing (pragma)

(3) Stoics (ca. 250 a.c.) lsignl semarnon semalnomenon,lekt6n

object or event

(4) Boethius (ca. 500) [wordl voice concepr thing

(5) Bacon (1605) [word] word nonon thing

(6) Leibniz (ca. 1700) [sign] sign character concepr thing

(7) Peirce srgn reDresentamen interpretant object

(B) Husserl (I900) qlm-'b^' exPresslon meanlng thing

(9) Ogden 6z Richards(1e23)

symbol thought orreference

referent

(10) Morns srSn sign vehicle significatum denotaum

Fig. Si 3. Synopsis of triadic modeis of the sign. (See also the synopsis of Meaning (Fig. M 3) and History ofSemiotics.)

90 I SIGN

lekton." Cf. Robins (1967: 16): "The Stoics

lormalized the dichotomy benveen form and

meaning, distinguishing in language 'the signi-

fier' and'the signified,'in terms strikingly rem-

iniscent of de Saussur e's signifant and signift'"

See also Robins (195I: 26ff ') and Eco (I984b:

29-33) and History (2.I'3).

4. Boethius: see 2.I.1: voice-vox' concept-

conceptio, thing-res.

5. Bacon: see Universal Language (2'2'I)'

6. Leibniz; see Poser (I979): sign-signum or

char ac ter, concept-co?lc ep tio (also no tio, c o glta-

tio, idea), thing-res.7. See Peirce.8. Husserl; see Meaning (3.I 3)9. Ogden & Richards (1923: II; cf' Lieb

I98Ia).I0. See Morris.

3. TRIADIC MODELS OF THE SIGN I 91

HANDBOOK OFSEMIOTICS

WINFRIED NOTH

"This is tlre most systematic discussion of semiotics yct published." -4hoice

"A bravura performance. . . ." -Thomas Sebeok

"Mapping a discipline as new as semiotics is more than an exercise in compilation anddescription. It involves substantive choices regarding the taxonomy and dimensions ofthe field as a whole. Noth succeeds admirably in this formidable undertaking."

-Librnry Joarnol

"Ncith's handbook is an outstanding encyclopedia that provides first-rate informationon many facets of sign-related studies, research results, and applications."

-Socisl Sciences in General

"Noth's book is a remarkable survey ofthe incredibly broad field of semiotics. . . . Forthose who desire a comprehensive survey of this diverse field. Noth's book is unsur-passed." -Library Qlorterly

Noth's ambitious work covcrs the field of semiotics from Aesthetics to Zoosemiotics.The book is organized in eight parts: "History and Classics of Modern Semiotics";"Sign and Meaning"; 'Semiosis, Codc, and the Semiotic Field"; "Ianguage and Lan-guage-Based Codes"; "From Suucturalism to Text Semiotics: Schools and Major Fig-ures"; "Tcxt Semiotics: The Field"; "Nonverbal Communication'; and "Aestheticsand Visual Communication." It contains sixty-five encyclopedic articles, a consoli-datcd bibliography of almost 3,000 tides, an index of names, and an index of terms.

Winfried Noth is Professor of English Linguistics at the University of Xhssel, Ger-many. N6th is the author of several books and numerous articles in the fields of lin-guistics and the semiotics of art, literature, and the media.

ADVANCBS IN SEMIOTICSThomas A. Sebeok. General Editor

INDIANA TINT\ZERSTIY PRES.SBLOOMINGTON & INDIANAPOLIS

rsBN 0-2s3-20959-5

, Uilil]ruilffiill rr nrfrrlrfrrl